Passive Tissues Help the Back Muscles to Generate Extensor Moments During Lifting

0021-9290(93)EOO12-E

PASSIVE TISSUES HELP THE BACK MUSCLES TO GENERATE EXTENSOR MOMENTS DURING LIFTING

P. DOLAN, A. F. MANNION and M. A. ADAMS Comparative Orthopaedic Research Unit, University of Bristol, U.K.

Abstract-We examined the possibility that passive tissues can help the erector spinae to generate large extensor moments during lifting. One hundred and forty-nine healthy men and women participated in the study. Subjects pulled upwards with steadily increasing force on a floor-mounted load cell, while EMG activity was recorded from electrodes overlying the erector spinae at L3 and TlO. Extensor moment was calculated from the load cell data, and was plotted against the full-wave rectified and averaged EMG signal. The relationship was linear with an intercept on the extensor moment axis (I) which indicated the flexion moment resisted by ‘passive (electrically silent) tissues. The dependence of I on lumbar flexion angle was studied by repeating the isometric pulls between 6 and 12 times, with the subject positioned in varying amounts of flexion, as measured by the ‘3-Space Isotrak’. Subjects then lifted weights of up to 20 kg from the floor, using ‘stoop’, ‘squat’ and ‘freestyle’ techniques, while lumbar flexion and EMG activity were recorded at 28 Hz. The isometric pulls showed that, on average, I increased from 25 Nm in lordotic postures to 120 Nm (for men) and 77 Nm (for women), in full flexion. During the lifts, peak extensor moment was generated with the lumbar spine flexed by 78-97% of the range between erect standing and full flexion. Comparisons with the static calibrations showed that between 16 and 31% of the peak extensor moment generated during lifting was unrelated to EMG activity in the erector spinae. Comparisons with cadaveric data suggested that less than a quarter of this ‘passive’ extensor moment was due to the intervertebral discs and ligaments.

INTRODUCTION the lumbar extensor muscles alone (McNeil1 et al., 1980). The thoracic erector spinae can generate an In stooped postures, the electrical activity of the extensor moment about the lumbar spine by means of erector spinae muscles falls to zero leaving the forward long tendons lying just underneath the lumbo-dorsal bending moment of the upper body to be resisted fascia (Bogduk et al., 1992; McGill and Norman, 1987) entirely by passive (electrically silent) structures. This but even so, there is barely enough active extensor ‘flexion-relaxation’ phenomenon depends on lumbar strength in the muscles (McGill et al., 1988). flexion rather than the overall angle of the trunk It appears likely that the muscles do receive assist- (Andersson et al., 1976; Floyd and Silver, 1955; Kip- ance from passive tissues, and the extent and origin of pers and Parker, 1984) and it can persist even when this assistance is of practical importance. If a sub- substantial weights are held in the hands (Floyd and stantial extensor moment is supplied by a structure Silver, 1955; Kippers and Parker, 1984; Schultz et al., lying just underneath the skin (for example, the lum- 1985). The origin of the ‘passive’ extensor moment is bo-dorsal fascia) then that structure has the advantage unknown, but it may involve the intervertebral disc ofa long lever arm about the ‘pivot’in the centre of the and ligaments, the iliolumbar ligaments, the lumbo- disc (Pearcy and Bogduk, 1988) and the spine will be dorsal fascia, collagenous tissue within the muscles subjected to a smaller compressive force than if the themselves, and a raised intra-abdominal pressure. same extensor moment had come from the muscles. The ‘passive’ moment may also involve remote mus- Similarly, a substantial extensor moment from the cles pulling actively on these structures. intervertebral ligaments would have a high ‘cost’ in Whatever its origin, a passive extensor moment has terms of lumbar compression because they lie closer to the potential to assist the muscles of the lumbar spine the pivot than the muscles. Therefore, calculations of during manual handling. When heavy weights are spinal compression during manual handling are re- lifted from the ground, the lumbar spine must generate liant on accurate information concerning the origins a high extensor moment in order to raise the upper of the extensor moment. body and weight into the upright position. Calcu- Another reason for wanting to apportion the exten- lations based on muscle cross-sectional area suggest sor moment between active and passive tissues is the that the required forces are beyond the capability of likelihood that they will respond differently to repetitive or chronic loading: metabolically active muscle will fatigue (Roy et al., 1989) and passive tissues will ‘creep’ (McGill and Brown, 1992). The highest risk Accepted in final form 16 September 1993. Address correspondence to: Dr P. Dolan, Comparative factor so far reported for acute disc prolapse is frequent Orthopaedic Research Unit, Department of Anatomy, Uni- bending and lifting (Kelsey et al., 1984) but the mech- versity of Bristol, Park Row, Bristol BSl 5LS, U.K. anisms leading to this association cannot be satis-

1077 1078 P. DOLANet al. factorily explored until the role of active and passive bodymass, height, mobility, and strength in lifting (see tissues in lifting tasks is better understood. below). Gracovetsky and Farfan have suggested several mechanisms which might lead to extensor moment generation from the lumbo-dorsal fascia (Gracovetsky A simple analysis of moments and Farfan, 1986; Gracovetsky et al., 1981, 1985) but A ‘moment arm analysis’ of the forces and moments their theories and calculations lack experimental veri- generated during lifting is shown in Fig. 1. The fication (Macintosh et al., 1987; McGill and Norman, 1988). McGill and colleagues (1988) have suggested that passive tissue involvement in heavy lifting is not essential and that, in practice, its contribution is small (Potvin et al., 1991). However, the predictions of their mathematical model depend greatly on an assumed function relating flexion angle and passive tissue moment, and this has lead to some inconsistent results (Potvin et al., 1991). The objectives of the present study were to measure the passive extensor moment during everyday bending and lifting activities, and to indicate its likely origin.

MATERIALSAND METHODS

Experimental design Extensor moment generation was calibrated against EMG activity in the erector spinae muscles during a series of isometric pulls, each performed with a different amount of lumbar flexion. These calib- ISOMETRIC CALIBRATION rations showed how the passive component of exten- OF E.M.G. SIGNAL sor moment depended on lumbar flexion. Then, lumbar flexion was measured continuously during dynamic lifting activities, using the ‘3-Space Isotrak device, and the peak flexion angles used to calculate the passive extensor moment during each lift. A large- scale study was undertaken to overcome the random errors inherent in skin-surface EMG measurements.

Subjects participating in the study intercept = I One hundred and forty nine healthy men and E.M.G. women volunteered for this study. None had any +, ACTIVITY history of severe low back pain. Most were nurses from Eo lmicrovolts) local hospitals, and had previously received some EM = W”D+w*dw = Eo*G+I training in how to lift. The other volunteers were white collar workers at the University and local hospitals. Fig. 1. Upper. A simple sagittal-plane model of forces and moments acting on the lumbar spine when the subject pulls Informed consent was obtained, but the objectives of upwards with force W. Lower. There is a linear relationship the study were not revealed and lifting technique was between extensor moment (EM) and EMG activity from the not discussed. Table 1 gives details of the subjects’ age, erector spinae muscles (Eo).

Table 1. Details of the 149 subjects. ROF=range of flexion, and EM, =maximum extensor moment generated during the isometric pull

Women (n = 126) Men (n = 23)

Mean STD Range Mean STD Range

Age (yr) 27.6 5.8 18-45 30.6 5.3 22-39 Bodymass (kg) 61.9 8.1 46.8-84.2 77.7 9.2 62.4-99.5 Height (cm) 165 6 150-183 179 I 167-193 Lumbar ROF (deg) 56.8 8.9 33.1-74.8 53.3 1.7 38.5-69.4 EM,,, (Nm) 245 51 145-383 430 116 212-640 Extensor moments during lifting 1079 assumptions underlying this model have been dis- During each pull, the curvature of the lumbar spine cussed previously (Dolan and Adams, 1993a). In static in the sagittal plane was measured at a frequency of equilibrium: 60 Hz using the 3-space Isotrak. The lever arms D and dw shown in Fig. 1 were estimated (Dolan and Adams, EM= WD+wdw, (1) 1993a) and input to the computer so that extensor where EM is the extensor moment, W, the vertical moment EM could be evaluated from the load cell force exerted on load cell and w, the weight of the data using equation (1). Extensor moment was then upper body and arms. D and dw are defined in Fig. 1. plotted against EMG activity (E,)as shown in Fig. 2. During an isometric contraction, EM is linearly (For ease of interpretation, every fifth data point was related to the EMG activity (E,) of the back muscles plotted so that the graph comprised of 40 points, (Dolan and Adams, 1993a) so that although all data were taken into account in subsequent analysis). Each subject performed between 6 EM=GEo+I, (2) and 12 calibrations with the lumbar spine ranging where G is the gradient of the graph and I is the between the slightly lordotic and fully flexed positions. intercept (see Fig. 1). About 3 min was allowed for recovery between succes- G and I can be obtained from calibrations as sive calibrations. described below, and so EM can be calculated from After a further recovery period, the subject stood measured values of E,. By definition, the active com- with the lumbar spine flexed by 70% and pulled up ponent of EM is zero when E,= 0,and so the intercept with maximum force for 3 s. Between two and four I indicates the passive extensor moment. repetitions were necessary to establish a consistent maximum extensor moment which could be used as an index of that subject’s strength (Table 1). It is inter- Isometric calibrations of EMG activity and esting to note that higher extensor moments could be extensor moment generated in more flexed postures, both in the static The full range of lumbar flexion was established for calibrations and during dynamic lifts. each subject by measuring lumbar curvature in the Additional experiments were performed on eight erect standing and extreme toe-touching positions subjects in order to clarify the extent of the EMG using the 3-Space Isotrak device. This consists of a source and sensor of pulsed electromagnetic waves which can be attached to the skin surface overlying the spinous processes of Ll and Sl in order to measure angular movements of the lumbar spine (Adams and Dolan, 199 1). Subjects then stood in a wooden frame with the front of their upper thighs resting against a cushioned crossbar, and with a strap fastened around their hips. Whilst leaning forwards in the frame, the subjects pulled upwards on a handlebar attached by a variable- length chain to a floor-mounted load cell, as shown in Fig. 1. The handlebar was constrained by the frame to move only in the vertical direction. As described 0 200 300 400 500 600 700 previously (Dolan and Adams, 1993a), subjects pulled LUMBAR EMG (pV) up on the handlebar with increasing force to reach a 0% ‘comfortable’ maximum in 3.3 s, whilst the output from the load cell was A-D converted at 60 Hz and input to a microcomputer. The electrical activity of the erector spinae muscles was measured by surface electrodes attached over the belly of the muscle on the left- hand side at T10 and L3. The location ofthe electrodes was chosen to optimise the linearity between extensor moment and EMG activity, and also to take account of tensile forces generated across the lumbar spine by thoracic regions of the erector spinae (Bogduk et al., k 1992). The EMG signal was full-wave rectified, aver- w aged with a time interval of 0.05 s, filtered with a band- IWW 450 600 pass of 5-300 Hz, A-D converted at 60 Hz and re- LUMBAR EMG @V) corded on the microcomputer. Subsequently, the Fig. 2. Isometric EMG-extensor moment calibration when EMG data were subjected to five-point smoothing the subject’s lumbar spine was (A)slightly flexed, and(B) fully and corrected for the effects of electromechanical flexed. In (A) the intercept, and hence the passive extensor delay. moment, was small, but in (B) it was large. 1080 P. DOLAN~ al. silence in full flexion. Extra electrode pairs were region was calculated using a least-squares algorithm. attached over the erector spinae at L3, on the right- The correlation coefficient (R) between I?, and exten- hand side, and over the latissumus dorsi (7 cm lateral sor moment was usually in the range 0.87-0.98, as to T7, at 45” to the long axis of the back). Calibrations described previously by ourselves (Dolan and Adams, were performed as described above. 1993a) and others (Seroussi and Pope, 1987). The intercept tended to be higher at L3 than at TlO, Spinal movements and loading during li@ing tasks especially in flexed postures. A passive extensor mo- Each subject performed five lifts. In the first three, a ment implies electrical silence in all regions of the pen (0 kg), a 10 kg weight-lifter’s disc and a 20 kg erector spinae, and so for each subject and each weight-lifter’s disc were lifted from the floor in what- calibration, the lower of the two intercepts was used as ever manner the subject preferred (freestyle). The the ‘passive’ extensor moment 1 in subsequent analyses. fourth and tifth lifts were standardised ‘squat’ and Additional EMG measurements on eight subjects ‘stoop’ lifts of the 10 kg disc, performed over 4 s to the showed that intercepts were similar bilaterally, and beat of a metronome. For the squat lift, the subject was tended to be greater for the latissimus dorsi, so I instructed to bend the knees, and in the stoop lift to represents true ‘flexion relaxation’ of the back muscles. keep them straight. Subjects were allowed to practice I increased markedly with lumbar flexion, and in the lifts before recordings were made so that they order to assess general trends, nine ranges of lumbar would become familiar with the effort required, and flexion were chosen, as listed in the first column of with the feel of the instruments attached to their back. Table 2. Values of Z and ‘percentage flexion’ were At least one minute was allowed for recovery between averaged using data from all subjects for all pulls that lifts, and no subjects showed any signs of fatigue. Eight fell within each range of flexion. Data for men and subjects performed an additional lift of the 10 kg disc women were analysed separately and the results are after being instructed to ‘keep as much lordosis as given in Table 2. Figure 3 shows how the ‘passive’ possible’. extensor moment increased with lumbar flexion. From During the lifts, lumbar curvature (LC) and the the lordotic standing position up to 80% flexion, the full-wave rectified and averaged EMG activity were increase is slight, but above 80% it rises considerably sampled at 28 Hz. Lumbar flexion was calculated for to about 120 Nm for men, and 77 Nm for women. each interval of l/28 of a second and expressed as a The women showed a weak but highly significant percentage of the subject’s full range of movement relationship between I and body mass (r’=0.053, using the formula: Percentage flexion = 100 x [LC - LC(standing)]/ [LC(full flexionbLC(standing)]. 90 80 (3) 70 The overall extensor moment (EM) was calculated r 60 from the EMG activity, using data from the isometric z - 50 calibrations on the same subject (Dolan and Adams, 1993a). Essentially, linear or polynomial curve fitting 40 was used to obtain expressions for G and I (Fig. 1) in 30 terms of lumbar curvature, and then equation (2) was 20 20 40 60 80 100 120 used to calculate extensor moment. EMG data were corrected for the electromechanical delay, and for the LUMBAR FLEXION (%) effects of muscle contraction velocity on the EMG/extensor moment relationship (Dolan and Adams, 1993a). Comparisons between lifts were made using a re- peated measures ANOVA. Significant differences are quoted at the 1% level unless stated otherwise. 100.

RESULTS

Isometric calibrations The EMG-extensor moment relationship was es- -l 20 40 60 80 100 120 sentially linear, as shown in Fig. 2. In flexed postures, the linear region was preceded by a short steep region LUMBAR FLEXION (%) typified by the trace in Fig. 2(B). The start of the linear Fig. 3. The ‘passive’ extensor moment I increased with region was found by visual inspection of the computer lumbar flexion for both men and women. Bars indicate the monitor, and the gradient and intercept of the linear SEM. Extensor moments during lifting 1081

Table 2. In the isomet ‘c pulls the intercept (I) of the extensor moment-EMG relationship depended on lum ar flexion. Values shown are the mean (SEM) for all pulls within “’ the specified flexion range Women Men

Average Average Flexion (%) flexion (%) (Am) flexion (%) (N*m)

<40 31.7 27.4 (3.9) 33.4 19.8 (9.7) 40-50 46.0 28.1 (2.6) 46.2 20.0 (4.9) 50-60 55.4 30.5 (2.4) 56.4 33.8 (6.0) 60-70 64.7 35.5 (1.9) 64.6 42.7 (6.7) 70-80 75.0 39.7 (2.2) 74.5 43.3 (6.4) 80-90 85.2 48.6 (2.1) 84.2 31.8 (9.7) 90-95 92.1 59.6 (2.8) 92.0 72.3 (9.6) 95-100 97.0 71.0 (2.5) 97.3 84.0 (8.3) >lOO 102.4 76.7 (3.4) 102.4 120.5 (9.5)

P < 0.0001; using pooled data for all flexion angles) but men, and equations (4) and (5) were used to calculate this was not apparent for the men (rZ=0.0003, P the passive contribution to the overall extensor mo- =0.823). Stepwise multiple regression showed that ment (using the average value of bodymass for women none of the other anthropometric variables accounted given in Table 1). The five lifts are compared in for any further variation in I which was therefore best Table 3. In the 10 kg lifts, stoop lifting required signi- described by the following equations: ficantly more flexion than squat or freestyle lifting and consequently generated a higher ‘passive’ extensor Women: I=491 x lo-” x F3+0.7573 x B-24.8 moment. The 10 kg freestyle lift, however, generated a (R2=0.321, p

Table 3. Peak values of extensor moment (EM) for the five lifts (f STD). Lumbar flexion and the passive extensor moment resisted (I) were evaluated at the point of peak extensor moment. Peak flexion at anytime during the lift is also shown

stoop Squat Freestyle lifts

10kg 10 kg Okg 1Okg 20 kg

Women Peak EM (Nm) 210+61 226+61 164+60 244+65 311&-83 Flexion (%) 96;tlO 84kl5 81rf:14 85_+15 90_+13 1 (Nm) 65.8 51.6 47.8 52.6 58.2 I (percentage of peak EM) 31.3 22.7 29.2 21.6 18.8 Peak Flexion (%) 100+8 87k14 83+12 88_+12 93_+11 Men Peak EM (Nm) 276_flll 278585 209_+20 328 + 109 384+ 140 Flexion (%) 97*9 81&16 78k12 79+16 85+12 1 (Nm) 85.0 55.6 50.1 52.8 61.3 I (percentage of peak EM) 30.8 20.0 24.0 16.1 16.0 Peak Flexion (%) 100+8 82+17 79_+12 82_+15 865 12 1082 P. DOLAN et al.

DISCUSSION stretched, electrically silent muscles themselves. The relative contributions of active and passive tissues to The size of the extensor moment resisted by passive muscle force production is not known, but cadaveric tissues depended very much on lumbar flexion, so the rectus abdominis muscle has a tensile strength of validity of the Isotrak measurements must be con- about 14 Ncm-* after rigor mortis subsides, and this, sidered. As discussed previously, these measurements presumably, can be attributed to non-contractile tis- can be accurate if the device is mounted on the skin sue (Katake, 1961). If the same figure is applicable to correctly (Dolan and Adams, 1993a) and the average the erector spinae, then the lumbar and thoracic range of flexion of our subjects (Table 1) is in excellent regions would contribute 66 Nm between them (calcu- agreement with mobility data obtained from radio- lated from McGill et al., 1988). This gives a total graphs (Adams and Hutton, 1982; Dvorak et al., 1991; passive extensor moment of 167 Nm (25+20+56 Pearcy et al., 1984). The reproducibility of Isotrak + 66) which comfortably exceeds our average value of measurements varies from about +2” in erect stand- 99 Nm. Several subjects recorded passive extensor ing to + 1” in full flexion (Adams and Dolan, 1991). It moments approaching 167 Nm, but these were parti- should be appreciated that the inferred passive exten- cularly strong men who should not be compared with sor moments generated during dynamic movements average cadaveric data. are insensitive to any systematic errors in flexion The contribution of passive tissues may not be angles because lumbar flexion serves only to com- entirely divorced from muscle activity. The isometric pare isometric calibrations with dynamic lifts, and calibrations in flexed postures showed that extensor the Isotrak was used to measure flexion on both moment could increase substantially, without any occasions. EMG activity from the erector spinae (Fig. 2(B)) and The origins of the passive extensor moment I are without any significant change in lumbar curvature. difficult to determine but some inferences may be This increasing extensor moment must have been drawn from the nature of its dependence on lumbar generated by muscles remote from our electrodes, and flexion (Fig. 3). In lordotic postures I does not fall then transmitted across the lumbar spine by passive much below 25 Nm even though there can be little tissues. Skin-surface electrodes have a relatively wide tension in ligaments and fascia lying posterior to the pick-up area and depth, and would detect strong centre of rotation. This 25 Nm may possibly be due to activation of muscles such as multifidus, as well as the a raised intra-abdominal pressure. In flexed postures, main erector spinae mass (Floyd and Silver, 1955). If I increases markedly as the limit of movement is the spine is sufficiently flexed, then ‘flexion relaxation’ approached, suggesting that passive structures poste- extends to the latissimus dorsi, psoas major, and rior to the spine are being brought into tension, after mutlifidus (Floyd and Silver, 1955), so the muscles being slack initially. responsible for this covert action may be the gluteals. It is of interest to compare I in fully flexed postures The action of the abdominal muscles pulling laterally with the known tensile strength of these posterior on the lumbo-dorsal fascia could make a minor contri- structures, and since the available strength data most- bution to I (Macintosh et al., 1987) although their ly refers to cadaveric material from both sexes, we will main function may be to prevent lateral contraction of consider the average value of I for men and women in the lumbo-dorsal fascia (Gracovetsky and Farfan, the highest flexion range, which is 99 Nm (Table 2). If 1986) thereby enhancing the action ofthe gluteals. The we suppose that a raised intra-abdominal pressure relationship between I and lumbar flexion is approx- contributes about 25 Nm in flexed postures also, then imately bi-linear (Fig. 3, upper) suggesting that two there is a further 74 Nm of passive extensor moment separate mechanisms may contribute to the passive to be accounted for. The intervertebral discs and extensor moment in different postures. ligaments can resist about 60 Nm when flexed to their In the analysis of the dynamic lifting tasks (Table 3) elastic limit (Adams and Dolan, 1991), but the protec- the dynamic flexion angle served as the point of tive action of the back muscles would normally limit comparison with the static calibrations. This proced- their contribution to about 20 Nm in static full flexion ure is justified by our previous finding that lumbar (Adams and Dolan, 1991). This leaves 54 Nm of I to be curvature (i.e. flexion angle) is the major determinant accounted for. Mechanical tests on the posterior layer of the EMG*xtensor moment relationship, and that of the lumbo-dorsal fascia (Tesh et al., 1987) and factors such as the knee angle, trunk angle and the supraspinous ligament (Myklebust et al., 1988) suggest height of the hands above the floor have little further that each of them could sustain a longitudinal force of influence (Dolan and Adams, 1993a). However, the about 330 N in the lower lumbar spine. Since these lie passive extensor moment may be increased in rapid about 9 and 8 cm posterior to the pivot (Tracy et a[., lifting movements, because collagenous tissue is visco- 1989), they could contribute a total of 56 Nm of elastic and resists rapid deformations more strongly extensor moment (330 N x 0.09 m + 330 N x 0.08 m). than slow ones. There is little quantitative data avail- This is just sufficient to supply the rest of the passive able to correct for this, but results from our own extensor moment but implies a very small margin of laboratory show that a motion segment’s resistance to safety. However, there is another potential source of 1 bending idcreases by 8% if the duration of the loading in addition to those already considered, and that is the cycle is decreased from 10 to 3 s, and by a further 2% if Extensor moments during lifting 1083 the loading cycle lasts only 1 s (Adams and Dolan; We have shown how nurses and white collar work- 1994). Conversely, visco-elastic effects would cause the ers lift moderately heavy weights off the floor, but this ‘passive’ extensor moments to be reduced during may not necessarily be the best way for them to lift. sustained or repetitive loading. Our own data suggest The high incidence of back injuries among nurses that the bending moment on the spine would fall by (Videman et al., 1984) prompts the question, should 17% after 100 full flexion movements performed over they be doing it differently? Our subjects found it 5 min, and by 42% after 5 min of sustained full flexion impossible to reduce lumbar flexion below about (Adams and Dolan, 1994). 57%, and it may not be wise for them to do so because The results for stoop and squat lifting could be lordotic postures generate high stress concentrations criticised on the grounds that the subjects were con- in the apophyseal joints (Adams and Hutton, 1980; strained to lift in a particular manner. However, this Dunlop et al., 1984) and posterior annulus fibrosus does not apply to the ‘freestyle’ lifts, and these yielded (Adams et al., 1993, 1994; McNally and Adams, similar results. Since the lifts were not performed in 1992). On the other hand, flexing beyond the normal random order, it is possible that some learning effect, static limit (100%) would increase the risk of ligament or fatigue, might influence the comparison between injury (Adams et al., 1980) and leave the posterior different lifts. Any such influence is unlikely to be annulus vulnerable to prolapse if high compressive large, however, because the subjects were allowed to forces were present (Adams and Hutton, 1982, 1985; practice the lifts beforehand, and an adequate re- Gordon et al., 1991). There remains a considerable covery period was allowed between lifts. On average, range of movement in which the lumbar spine could be our subjects flexed their lumbar spine by 81% in the positioned, and the question now is: which part of this 0 kg lift, rising to 89% in the 20 kg lift, even though range, if any, is to be preferred? most of them had previously received instruction on There are two schools of thought. McGill and co- lifting technique, and were careful to bend their knees. workers advocate lordosis because it allows the back Many individuals exceeded lOO%, and this is not a muscles to resist much of the forward shear force contradiction in terms: 100% refers to the static toe- acting on the spine, and because it reduces the tensile touching posture, and static limits can be exceeded forces in the intervertebral ligaments which lie close to during dynamic movements because the gravitational the centre of sagittal movement and therefore impose forward bending moment is augmented by the for- a high compressive ‘penalty’ on the discs (Potvin et al., ward angular momentum of the trunk. Similar flexion 1991). The need to reduce shear forces has not been angles in lifting have been measured previously using a established, since the lower lumbar apophyseal joints variety of techniques (Adams and Dolan, 1991; Davis are able to resist 1.2-2.5 kN before fracture (Cyron et al., 1965; Dolan and Adams, 1993b; Potvin et al., et al., 1976) but the second benefit is readily apparent. 1991). None of our eight subjects found it possible to Conversely, Gracovetsky and colleagues (Gracovet- lift a weight from the ground without substantially sky and Farfan, 1986; Gracovetsky et al., 1981, 1985) flexing their lumbar spine, and the frequently offered appear to advocate full flexion because it enables the advice to ‘keep the lumbar lordosis when lifting’ lumbo-dorsal fascia to resist more of the flexion appears to be based upon unreliable visual estimates moment. This structure lies about 9 cm posterior to of spinal posture. Perhaps a slight concavity in the the centre of rotation (Tracy et al., 1989) and so region T9-Ll has been confused with a true lumbar imposes a small compressive penalty on the discs. lordosis? Also, tension in the middle layer of the fascia increases the stability of the lumbar spine in the frontal plane (Tesh et al., 1987), and the energy-storing potential of the fascia can reduce the metabolic cost of lifting + c’ 100 weights up off the floor (Gracovetsky and Farfan, M -c- 1986). 5z, 80 We suggest that this debate has become polarised because different sources of ‘passive’ extensor moment z 60 have not been considered separately. Figure 4 suggests “0 40 0 l that it is possible to benefit from a high ‘passive’ ? $I 20 extensor moment, and yet avoid high tensile forces in those structures which lie closest to the centre of r IzzL! 040 00 80 100 120 rotation. The data concerning the intervertebral discs LUMBAR FLEXION (%) and ligaments is from a previous study on cadaveric lumbar ‘motion segments’ (Adams and Dolan, 1991) Fig. 4. The upper curve shows that the passive extensor and it shows that their resistance to bending (M) moment I increases markedly when lumbar flexion exceeds remains relatively low in the physiological (in vim) 80% of the in viuo range of movement (data from Table 2, averaged for men and women). In the range 80-95% flexion, range of movement. Most ofthe subjects in the present less than 25% of I is attributable to the intervertebral discs study flexed their lumbar spines by 80-95% during and ligaments. This component of I is denoted M (data from the lifts, and in this range of movement, M contributes Adams and Dolan, 1991). less than 25% of the total passive extensor moment I. 1084 P. DOLAN et al.

CONCLUSIONS Dolan, P. and Adams, M. A. (1993b) The influence of lumbar and hip mobility on the bending stresses acting on the (1) When lifting weights from the ground, most lumbar spine. C/in. Biomechanics 8, 185-192. people flex their lumbar spine by about 80-95% at the Dunlop, R. B., Adams, M. A. and Hutton, W. C. (1984) Disc space narrowing and the lumbar facet joints. J. Bone Jt time when the extensor moment is greatest. Surg. t&i-B, 706-710. (2) During stoop and squat lifts of 10 kg, about 31 Dvorak, J., Panjabi, M. M., Chang, D. G., Theiler, R. and and 21%, respectively, of the total extensor moment is Grob, D. (1991) Functional radiographic diagnosis of the unrelated to EMG activity in the erector spinae. lumbar spine. Flexion-extension and lateral bending. Spine (3) Less than 25% of this ‘passive’ extensor moment 16, 562-571. Floyd, W. F. and Silver, P. H. S. (1955) The function of the comes from the intervertebral discs and ligaments. The erectores spinae muscles in certain movements and pos- rest probably comes from the lumbodorsal fascia, the tures in man. J. Physiol. 129, 184-203. supraspinous ligament, non-contractile tissue within Gordon. S. J.. Yantz. K. H.. Maver. P. J.. Mace. A. H.. Kish. V. the erector spinae muscles, and a raised intra-abdom- L. ani Rabin, g L. (1691) hebhanism of ‘disc rupture- a preliminary report. Spine 16, 450-456. inal pressure. Gracovetsky, S. and Farfan, H. (1986) The optimum spine. (4) Muscles remote from the lumbar spine are able Spine 11, 543-571. to increase the passive extensor moment without Gracovetsky, S., Farfan, H. and Helleur, C. (1985) The increasing lumbar flexion. abdominal mechanism. Spine 10, 317-324. Gracovetsky, S., Farfan, H. F. and Lamy, C. (1981) The mechanism of the lumbar spine. Spine 6,249-262. Acknowieduements-This work was funded by the Medical Katake, K. (1961) Studies on the strength of human skeletal Research Council and the Arthritis and Rheumatism Council muscles. j. Kyoto Pref: Med. Univ. 69, 463-483. of Great Britain. We would like to thank Sheila James. Julie Kelsev. J. L.. Githens. P. B.. White. A. A.. Holford. T. R. et al. Milkins and Chris Standell for helping with these experi- (19&l) Ankpidemidlogic $tudy oklifting and twiiting on the ments. job and risk for acute prolapsed lumbar intervertebral disc. J. Orthop. Res. 2, 61-66. Kippers, V. and Parker, A. W. (1984) Posture related to REFERENCES myoelectric silence of erectores spinae during trunk flexion. Spine 9, 740-745. Adams, M. A. and Dolan, P. (1991) A technique for quanti- McGill, S. M. and Brown, S. (1992) Creep response of the fying bending moment acting on the lumbar spine in vivo. lumbar spine to prolonged full flexion. C/in. Biomechanics J. Biomechanics 24, 117-126. 7, 43-46. Adams, M. A. and Dolan, P. (1994) Time-dependent changes McGill, S. M. and Norman, R. W. (1987) Effects of an in the lumbar spine’s resistance to bending. Clin. Bio- anatomically detailed erector spinae model on LA/L5 disc mechanics (in press). compression and shear. J. Biomechanics 20, 591-600. Adams, M. A. and Hutton, W. C. (1980) The effect of posture McGill, S. M. and Norman, R. W. (1988) Potential of on the role of the apophyseal joints in resisting inter- lumbodorsal fascia forces to generate back extension vertebral compressive force. J. Bone Jt Surg. 62-B, moments during squat lifts. J. Biomed. Engng 10,312-318. 358-362. McGill, S. M., Patt, N. and Norman, R. W. (1988) Measure- Adams, M. A. and Hutton, W. C. (1982) Prolapsed inter- ment of the trunk musculature of active males using CT vertebral disc. A hyperllexion injury. Spine 7, i84-191. scan radiography: implications for force and moment Adams. M. A. and Hutton. W. C. (1985) Gradual disc generating capacity about the L4/L5 joint. J. Biomechanics prolapse. Spine 10.524-531. . ’ 21, 329-341. Adams, M. A., Hutton, W. C. and Stott, J. R. R. (1980) The Macintosh, J. E., Bogduk, N. and Gracovetsky, S. (1987) The resistance to flexion of the lumbar intervertebral joint. biomechanics of the thoracolumbar fascia. Clin. Bio- Spine 5, 245-253. mechanics 2, 78-83. Adams, M. A., McNally, D. S., Chinn, H. and Dolan, P. McNally, D. S. and Adams, M. A. (1992) Internal inter- (1994) Posture and the compressive strength of the lum- vertebral disc mechanics as revealed by stress profilometry. bar spine? International Society of Biomechanics Award, Spine 17(l), 66-73. 1993. Clin. Biomechanics (in press). McNeill, T., Warwick, D., Andersson, G. and Schultz, A. Adams, M. A., McNally, D. S. Wagstaff, J. and Goodship, (1980) Trunk strength in attempted flexion, extension and A. E. (1993) Abnormal stress concentrations in lumbar lateral bending in healthy subjects and patients with low- intervertebral discs following damage to the vertebral back disorders. Spine 5,529-537. body: a cause of disc failure. European Spine Society Myklebust, J. B., Pintar, F., Yoganandan, N. and Cusick, J. F. (Acromed) Award, 1992. Eur. Spine J. 1,214-221. et al. (1988) Tensile strength of spinal ligaments. Spine 13, Andersson, G. B. J., Herberts, P. and Ortengren, R. (1976) 526-531. Myoelectric back muscle activity in standardised lifting Pearcy, M. J. and Bogduk, N. (1988) Instantaneous axes of postures. In International Series on Biomechanics, Vol. 1A. rotation of the lumbar intervertebral joints. Spine 13, Biomechanics V-A (Edited by P. V. Komi) University Park 1033-1041. Press, Baltimore, pp. 520-529. Pearcy, M., Portek, I. and Shepherd, J. (1984) Three-dimen- Bogduk, N., Macintosh, J. E. and Pearcy, M. J. (1992) A sional X-ray analysis of normal movement in the lumbar universal model of the lumbar back muscles in the upright spine. Spine 9, 294-297. position. Spine 17, 897-913. Potvin, J. R., McGill, S. M. and Norman, R. W. (1991) Trunk Cvron. B. M., Hutton, W. C. and Troup, J. D. G. (1976) muscle and lumbar ligament contributions to dynamic lifts -Spohdyloly;ic fractures. .I. Bone Jt St& 58-B, 462-466. with varying degrees of trunk flexion. Spine 16,1099-l 108. Davis. P. R.. Trouo. J. D. G. and Bumard, J. H. (1965) Rissanen, P. M. (1960) The surgical anatomy and pathology Movemen& of thd ihoracic and lumbar spine when lifting: of the supraspinous and interspinous ligaments of the a chrono-cyclophotographic study. J. Anat. 99(l), 13-26. lumbar spine with special reference to ligament ruptures. Dolan, P. and Adams, M. A. (1993a) The relationship Acta Orthop. Stand. (suppl 46). between EMG activity and extensor moment generation in Roy, S. H., De Luca, C. J. and Casavant, D. A. (1989) Lumbar the erector spinae muscles during bending and lifting muscle fatigue and chronic lower back pain. Spine 14, activities. J. Biomechanics 26, 513-522. 992-1001. Extensor moments during lifting 1085

Schultz, A. B., Haderspek-Grib, K., Sinkora, G. and War- abdominal muscles and vertebral stability. Spine 12, wick, D. N. (1985) Quantitative studies of the 501-508. flexion-relaxation phenomenon in the back muscles. J. Tracy, M. F., Gibson, M. J., Szypryt, E. P., Rutherford, A. Orthop. Res. 3(2), 189-197. and Corlett, E. N. (1989) The geometry of the muscles of Seroussi, R. E. and Pope, M. H. (1987) The relationship the lumbar spine determined by magnetic resonance imag- between trunk muscle electromyography and lifting mo- ing. Spine 14, 186-193. ments in the sagittal and frontal planes. J. Biomechanics Videman, T., Nurminen, M. and Tola, S. et al. (1984) Low- u)(2), 135-146. back pain in nurses and some loading factors of work. Tesh, K. M., Shaw Dunn, J. and Evans, J. H. (1987) The Spine 9, 209-213.